Citations, evaluated in some fashion to take into account the size of institution, are the best understood and most widely accepted measure of research strength.
Often calculated on a “per paper” basis, the QS World University Rankings™ has adopted a “per faculty member” approach since its inception in 2004. The Citations per Faculty score contributes 20% to the overall rankings score.
For the calculation of this indicator, QS gathers two distinct datasets:
Citations count for the last five years
There are three major sources of publication and citation data worldwide, these are the Web of Science from Thomson Reuters; Scopus from Elsevier and Google Scholar. In the first three years of the QS World University Rankings™, results from the Essential Science Indicators (ESI), a subset of the Web of Science were used. In 2007, the switch was made to Scopus for a number of reasons, but principally due to broader journal coverage leading to results for a larger number of institutions.
A key development in 2011 was the exclusion of self-citations.
In 2015, two significant additional modifications were applied:
- Papers featuring authors from more than ten affiliated institutions have been excluded – this represents around 0.34% of the database and prevents highly cited material produced by very large research groups conferring too much credit on institutions who have only contributed in very small part to the work. Whilst often high-profile and important research, these papers often cause a distortion for a university that may not be especially research active otherwise.
- Faculty Area Normalization – due to publishing patterns and practices, a straight ratio of citations per faculty places a strong emphasis on life sciences and medicine. In consultation with advisors and participants, QS has opted to adopt a model which aims to equalize the influence of research in our five key faculty areas. A more detailed technical explanation of how this works can be found here.
In 2016, a further modification was made to the first of these changes. Rather than a fixed cap of ten affiliated institutions, we have applied a cap that is sensitive to the publishing patterns of the discipline to which a given paper belongs. In the previous approach, a larger proportion of papers in Physics would have been excluded when compared to Civil Engineering, for example. The variable cap has been calibrated to ensure that no more than 0.1% of research are excluded from any discipline.
Full Time Equivalent (FTE) faculty
Faculty numbers used are totals… whilst it would be ideal to separate the notions of teaching and research and use the former for calculating the Student Faculty Ratio and the latter for this indicator, it has not been possible to do so as data to that degree of distinction has so far proved unavailable for many countries in the study. The definition of exactly what data we request has evolved gradually over the years to minimize ambiguity.